Advanced Biomass Fuel Characterization Based on Tests with a

*Telephone: +43/316/48130013. ... Within the scope of several projects, the lab-scale reactor system as well as the subsequent evaluation routines hav...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/EF

Advanced Biomass Fuel Characterization Based on Tests with a Specially Designed Lab-Scale Reactor Thomas Brunner,*,†,‡,§ Friedrich Biedermann,†,‡ Werner Kanzian,†,§ Nikola Evic,†,§ and Ingwald Obernberger†,‡,§ †

BIOENERGY 2020+ GmbH, Inffeldgasse 21b, 8010 Graz, Austria BIOS BIOENERGIESYSTEME GmbH, Inffeldgasse 21b, 8010 Graz, Austria § Institute for Process and Particle Engineering, Graz University of Technology, Inffeldgasse 13, 8010 Graz, Austria ‡

ABSTRACT: To examine relevant combustion characteristics of biomass fuels in grate combustion systems, a specially designed lab-scale reactor was developed. On the basis of tests performed with this reactor, information regarding the biomass decomposition behavior, the release of NOx precursor species, the release of ash-forming elements, and first indications concerning ash melting can be evaluated. Within the scope of several projects, the lab-scale reactor system as well as the subsequent evaluation routines have been optimized and tests with a considerable number of different biomass fuels have been performed. These tests comprised a wide variation of different fuels, including conventional wood fuels (beech, spruce, and softwood pellets), bark, wood from short rotation coppice (SRC) (poplar and willow), waste wood, torrefied softwood, agricultural biomass (straw, Miscanthus, maize cobs, and grass pellets), and peat and sewage sludge. The results from the lab-scale reactor tests show that the thermal decomposition behavior and the combustion behavior of different biomass fuels vary considerably. With regard to NOx precursors (NH3, HCN, NO, N2O, and NO2), NH3 and, for chemically untreated wood fuels, also HCN represent the dominant nitrogen species. The conversion rate from N in the fuel to N in NOx precursors varies between 20 and 95% depending upon the fuel and generally decreases with an increasing N content of the fuel. These results gained from the lab-scale reactor tests can be used to derive NOx precursor release models for subsequent computational fluid dynamics (CFD) NOx post-processing. The release of ash-forming vapors also considerably depends upon the fuel used. In general, more than 91% of Cl, more than 71% of S, 1−51% of K, and 1−50% of Na are released to the gas phase. From these data, the potential for aerosol emissions can be estimated, which varies between 18 mg/Nm3 (softwood pellets) and 320 mg/ Nm3 (straw) (dry flue gas at 13% O2). Moreover, these results also provide first indications regarding the deposit formation risks associated with a certain biomass fuel. In addition, a good correlation between visually determined ash sintering tendencies and the sintering temperatures of the different fuels (according to Ö NORM CEN/TS 15370-1) could be observed.

1. INTRODUCTION AND OBJECTIVES

procedures to convert the nitrogen compounds into elemental nitrogen instead of nitrogen oxides. To realize an advanced fuel evaluation tool, which tackles the problems mentioned above, a special lab-scale reactor has been designed and constructed at the Institute for Process and Particle Engineering, Graz University of Technology. The aim was to design a batch reactor, with which the combustion on a grate can be simulated and which can be applied to cover all fuel characterization aspects mentioned above. Therefore, test runs with this reactor should provide information regarding the biomass decomposition behavior, the release of NOx precursor species, the release of ash-forming elements, and first indications concerning ash melting. In this paper, first the idea behind this lab-scale reactor is explained and then the reactor itself is introduced. Furthermore, results from recently performed test runs with 10 different biomass fuels are presented and compared to each other as

The use of conventional wood fuels, new agricultural biomass fuels, biomass residues from different processes, and pretreated fuels is of increasing interest in all European Union (EU) countries because of the growing relevance of energy use from biomass. Moreover, fuel flexibility is of rising importance for utilities as well as biomass furnace and boiler manufacturers. Therefore, an improved characterization of fuels and fuel mixtures is gaining increasing relevance. In this respect, it is essential to gain knowledge about the combustion and volatiles release behavior of different fuels and fuel mixtures. This is of special relevance for “new” and “difficultto-use” fuels (e.g., short rotation crops, energy grasses, and residues from agricultural industries), which usually show considerably higher ash contents and lower first ash melting points in comparison to conventional wood fuels (wood pellets, woodchips, and bark), leading to increased problems concerning slagging, ash deposit formation, and fine particulate emissions. Moreover, these fuels may also contain other precursors of pollutants, such as nitrogen, as part of the organic structure that may be oxidized into nitrogen oxides. An improved knowledge of the transformations of nitrogen compounds released from the fuel and carried to the exhaust gas may form a basis for reduction © 2013 American Chemical Society

Special Issue: Impacts of Fuel Quality on Power Production and the Environment Received: March 29, 2013 Revised: July 22, 2013 Published: July 23, 2013 5691

dx.doi.org/10.1021/ef400559j | Energy Fuels 2013, 27, 5691−5698

Energy & Fuels

Article

Figure 1. Scheme of the lab-scale reactor, including measurement setup. B1, B2, and B3 indicated the three different height levels of the temperature measurements in the fuel bed. Dimensions are in millimeters. Flue gas samples are extracted from the gas volume above the fuel bed, partly treated, respectively conditioned (dilution and temperature stabilization), and introduced into the following gas analysers: (i) Fourier transform infrared spectroscopy (FTIR) (Ansyco), CO2, H2O, CO, CH4, NO, NH3, HCN, NO2, N2O, and basic hydrocarbons; (ii) Emerson NGA 2000; paramagnetism, O2; nondispersive infrared analysis (NDIR), CO2 and CO; heat conductivity, H2; (iii) flame ionization detector (FID) (Bernath Atomic 3005), OGC (CxHy); (iv) chemiluminescence detector (CLD) (ECO Physics CLD 700 EL ht), NO, NOx; and (v) wide band λ sensor, O2. Moreover, the flow and temperatures of the reaction medium (e.g., air, N2, or different mixtures of O2 and N2) and several flue gas and reactor temperatures are recorded with NiCr−Ni thermocouples. The testing protocol was defined as follows: (i) Before the test run, the fuel is pre-dried to 10 wt % wet basis (wb) moisture content and a sub-sample of the fuel is forwarded to wet chemical analyses. (ii) First, the fuel is filled into the sample holder and lightly compressed to gain a packed bed with a realistic density. (iii) Then, the reactor is preheated, applying 750 and 450 °C as pre-settings for the upper and lower heating element. These settings are kept constant during the whole test run. (iv) Then, the sample holder with the fuel is introduced into the reactor, and the reaction gas flow through the grate and the fuel bed is activated. All experiments documented in this paper have been performed with dry air (21 vol % O2 and 79 vol % N2) as the reaction agent and a gas flow of 30 LN/min. The reaction gas flow rate is kept constant during the whole test run. (v) All parameters mentioned above are continuously recorded in a 2 s interval over the whole test run, which usually lasts between 30 and 60 min depending upon the fuel mass applied. (vi) At the end of the test run, first the residues (ashes) are visually evaluated regarding ash sintering and slag formation and then removed and forwarded to chemical analyses. The lab-scale reactor has been designed to represent the burning conditions of a biomass fuel layer on a grate as good as possible (Figure 2). It reproduces the behavior of a fuel segment moving along the grate and, thereby, passing first through the drying zone, then through the devolatilization and charcoal gasification zone, and finally, through the char burnout zone. Because the air flow through the fuel bed remains the same during the whole experiment, the excess air ratios vary depending

examples for the application of the reactor as an efficient novel fuel characterization method.

2. METHODS 2.1. Description of the Lab-Scale Reactor. The basic idea was to develop, design, and construct a lab-scale batch reactor, which is capable for the simulation of the fuel decomposition behavior in real-scale fixedbed thermal biomass conversion systems. Therefore, the following constraints were given: (i) reasonable sample intake to consider secondary reactions in the fuel bed appropriately, (ii) high heating rates of the fuel comparable to real-scale grate furnaces, (iii) inert reactor material to avoid reactions of the gases with the reactor, (iv) high flexibility regarding analytical equipment connected with the reactor, and (v) online recording of relevant operation data and emissions as well as the mass loss. Finally, the following concept was realized (Figure 1): The core of the reactor is a cylindrical retort (height, 35 cm; inner diameter, 12 cm), which is heated electrically and controlled by two separated proportional− integral−derivative (PID) controllers (see Figure 1). The fuel is put in a cylindrical sample holder of 100 mm height and 95 mm inner diameter. This size provides the possibility to use a fuel mass between around 100 g (for low-density fuels, such as chopped or chipped fuels) and more than 400 g (for pelletized fuels), which is a reasonable quantity for the simulation of a packed bed. Moreover, the sample holder is equipped with five thermocouples (NiCr−Ni) to monitor the fuel bed temperatures during the test runs (see Figure 1). The material of the reactor wall and the sample holder is silicon carbide. This material had already proven its applicability for such types of reactors in former experimental setups and was selected because it is inert under reducing and oxidizing conditions and, therefore, does not react with the fuel, ash, and flue gas. The mounting and vessel for the fuel bed are placed on the plate of a scale. The scale is mechanically separated from the retort by a liquid sealing (synthetic thermal oil: Therminol 66) and is used to determine the weight loss of the sample over the test run period. The sample holder with the biomass is introduced into the preheated reactor, and therefore, a rapid heating, which is well comparable to the heating in real-scale thermal conversion processes, can be achieved. 5692

dx.doi.org/10.1021/ef400559j | Energy Fuels 2013, 27, 5691−5698

Energy & Fuels

Article

treated with acid and the generated CO2 is measured by infrared (IR). TOC is determined by the amount of total carbon (determined by an element analyzer) minus the TIC. For the determination of major and minor elements, expect Cl, multi-step pressurized digestion with HNO 3 (65%)/HF (40%)/H 3BO 3 followed by measurement by ICP−OES is applied. The Cl content of ashes is measured by ion chromatography after elution for 24 h with deionized water. 2.3. Evaluation of Lab-Scale Reactor Test Runs. In this section, the evaluation strategy of lab-scale rector test runs is briefly presented to provide a better basis for the understanding of section 3, where results from test runs with different biomass fuels are discussed. At first, a validity check regarding all data gained from the test runs has to be performed. Therefore, from the reaction gas flows measured, the weight loss recorded by the balance, and the results of the flue gas analyses, cumulative balances for the main elements C, H, and O are calculated. Moreover, using the results of the fuel analyses regarding the oxide-based ash content and the ash analyses regarding the TOC and TIC contents, an ash mass balance is calculated. Considering the inhomogeneous composition of biomass fuels and the possible measurement and analysis error ranges, a full closure of the mass and element balances is not realistic. However, experience has shown that recovery rates in the range between 80 and 120% are reasonable, and therefore, only test runs with recovery rates within these margins are further evaluated. The recovery rate is thereby calculated by dividing the mass of an element in the flue gas and the ashes with the mass of the respective element in the initial test run fuel. In a next step, characteristic data describing the thermal fuel decomposition are evaluated. These data include the start and duration of the volatilization and charcoal gasification phase and the main charcoal combustion phase and the maximum fuel bed temperatures measured and the decomposition rates determined during these phases. Then, the flue gas species and the trend of the excess air ratio over the whole test run duration are investigated. Special emphasis is thereby put on the trends regarding the release of NOx precursors. Finally, the data gained from the chemical analyses of the ashes are evaluated. In a first step, element balances for the major ash-forming refractory species (Ca, Si, Mg, Al, and Fe) based on the initial fuel sample mass (determined prior to the test run), the mass of the residual ash (determined after the end of the test run), and the results of the fuel and ash analyses are calculated as a plausibility check for the ash sampling and the analyses. It is important that these balances show a good closure with recovery rates between 80 and 120%. However, this balance closure is often not accurate enough to calculate release rates for, e.g., K, which can reach values